Hello,
It wasn't entirely clear to me what your actual difficulty is but I note
that you haven't received a reply yet so here's one.

>
> Each subject is prsented with an instruction, 'word' or 'colour' followed by
>the post instructions period then the stimulus is presented during scan 6 and
>the response(button press) occurs. Scans 1-5 are the instruction-related
scans,
>scans 6-10 the response-related scans. The stimulus is congruent (eg RED
in red)
>or incongruent( RED in purple). We have 480 scans, 48 trials therefore 10
>scans/trial. I have set 4conditons, word, colour, congruent, and incongruent.
>Word or Colour is presented 24 times in random order and stimulus
presentation
>congruent or Incongruent, presented 24 times in random order.  Am I
correct to
>do the following:

So the four conditions embody two factors - instructions (word vs colour)
and judgement (congruent versus incongruent). As far as I can see, your
setting up is perfectly correct. When asked for stimulus onsets (just after
you have specified SOA - variable), I am assuming that you are entering,
for each condition (or trial type) a 24-element vector which notes the
start scan for each of the 24 events of that type. If that is so, then I
don't see any reason why this wouldn't generate a nice design matrix with 4
columns per session and you would then be able to use the results button to
explore these in the normal way. That is, if the order of conditions is
[word, colour, congruent, incongruent]
then a contrast weighting of
[1 1 0 0]
would tell you about the brain regions showing a main effect of
instructions irresepective of instruction type
[1 -1 0 0]
about the effect of instructions that are word-specific
etc.


Added possibilities.
The use of the canonical HRF (and non-variable duration) models a rather
specific effect. At present, I am assuming that you are treating the onset
of each event as the presentation of the stimulus and, implicitly, treating
this as an instantaneous event. You might alternatively treat the
instruction period as a sustained period of pre-task preparation which is
better modelled as an event of duriation 5 scans. You might also introduce
greater flexibility by modelling the event with derivative and dispersion
to accommodate the possibility of different lags or duration of responses
respectively (although these additions do make things rather more
complicated when you are carrying out the pairwise comparisons at the next
stage).

One other point is that you could explore the interaction of the two
factors at the response stage by modelling 4 types of response rather than
two. Thus, a response event could be specified not just accoriding to its
congruity or incongruity but also according to the instruction that
preceded it. This would lead you to a design matrix with 6 event types (2 x
instruction as before and 4 x response).

Anyway, having presumably succeeded in complicating matters, I shall wish
you good luck

Paul Fletcher



-----------------------------------------------------------------
Paul Fletcher,
Research Department of Psychiatry,
University of Cambridge,
Addenbrooke's Hospital,
Hills Road,
Cambridge,
UK
CB2 2QQ

Tel 01223 336 988
Fax 01223 336 581